106 Journal of Medicinal Chemistr), 1977, Vol. 20, N o . 1
over each other for all the active isomers. This superimposibility is also shown by comparison of distances between key features of the proposed pharmacophore as shown in Table I. Since there are no low-energy 2-SS axial isomers, no 2-SS (or R R ) isomer can superimpose on this pattern. N o conformer of the 3 compounds can be matched to the pattern shown in Figures 4 and 5, thus possibly accounting for their activity. The active conformation of each molecule is such that it allows a hydrogen bond to form between the OH and amine functions. The existence of a hydrogen bond is not necessarily required for activity but helps to stabilize each active molecule in the conformation that contains the proposed pharmacophore. ChenglGhas also suggested that the axial conformation of compound 2 must be invoked to explain its stereospecific activity and uses mass spectral data to lend support to this conclusion. Our results, based on direct conformational analysis, independently describe the same pharmacophore to account for the stereospecificity of the 2 isomers, the lack of it in the 1 isomers, and the inactivity of the 3 isomers. They further point to the 2-SS and 2-RR as the inactive forms, a prediction that can be tested by experiment. Subject to further verification then, our results have led to a description of a pharmacophore for this subclass of antimalarial drugs which can account for their observed
Rolte, Demuynck, L h o m m e
behavior and is similar in dimension and character to that proposed by Cheng.
References and Notes (1) F. I. Carroll and J . T. Blackwell, J . Med. Chem., 17, 210 (1974). (2) P-L. Chien and C. C. Cheng, J . Med. Chem.. 16,1093 (1973). ( 3 ) R. E. Olsen. J . Med. Chem., 15, 207 (1972). (4) P-L. Chien, D. J. McCaustland, W. H. Burton, and C. C. Cheng, J . Med. Chem., 15, 28 (1972). (5) C. C. Cheng, personal communication. (6) E. E. Davies, D. C. Warhurst, and W. Peters, A n n . Trop. Med. Parasit.. 69, 147 (1975). ( 7 ) D. C. Warhurst, C. A. Homewood, W. Peters, and V. C. Baggaley, Proc. Helm. Soc., Wash., 39, 271 (1972). (8) C. D. Fitch, Proc. Helm. Soc., Wash., 39, 265 (1972). (9) D. C. Warhurst and P. Mallory, Trans. R. Soc. Trop. Med. Hyg., 67, 20 (1973). (10) C. C. Cheng, J . Pharm. Sci., 60, 1596 (1971). (11) S. Diner, J. P. Malrieu, F. Jordam, and M. Gilliert, Theor. Chim. Acta, 15, 100 (1969), and references cited therein. (12) J. Pople and D. Berveridge, “Approximate Molecular Orbital Theory”, McGraw-Hill, Kew York, N.Y., 1970. (13) D. C. Phillips. Acta Crystallogr., 7 , 159 (1954). (14) C. Rerat, Acta Crptallogr., 13, 72 (1960). (15) G. H. Loew and D. S. Berkowitz, J . Med. Chem., 18, 656 (1975). (16) C. C. Cheng, J . Med. Chem., 19: 170 (1976).
Synthetic Models of DNA Complexes with Antimalarial Compounds. 2. The Problem of Guanine Specificity in Chloroquine Binding J. Bolte, C. Demuynck, and J. Lhomme* Laboratoire d e C h i m e - d e s Substances Naturelles (E.R.A. d u C.N.R.S. N o . 392), Centre Unicersztaire des Sciences et Techniquea, C‘niLersite de Clermont, B.P. 45, 631 70 .4uhzere, France. Received Februar) 9, 1976
Stacking interactions between the aminoquinoline ring of the antimalarial chloroquine and the purine bases have been studied by preparing and examining models in which the quinoline is linked to the base by a trimethylene chain. The degree of stacking of the models which reflects the strength of the interaction was quantitatively determined in water at different temperatures by hypochromism measurement in the uv. Adenine and guanine exhibit equal affinity for the auinoline nucleus as reflected bv” verv“ close hypochromism values observed for the two models a t all temperatures. studied.
Chloroquine ( l ) , l a widely used antimalarial, inhibits several biological functions of plasmodia, the microorganism responsible for the disease, acting a t the level of the DNA and RNA polymerases and interfering with protein synthesis.’ Its mode of action involves binding to the nucleic acids; the influence of this on nucleic acid biosynthesis has been demonstrated in vitro.3 The complexation of chloroquine with DNA4 and with synthetic polynucleotides5 has been studied. Two types of interaction are involved in complex formation, an electrostatic attraction between the amine group (protonated at physiological pH6) of the chloroquine side chain and the phosphate groups of the DNA, and a more specific interaction between the aromatic system of chloroquine and the nucleotide bases.7 It has been suggested that the protonated aminoquinoline ring is actually intercalated between base pairs in the DNA helix.8 As an approach to the problem of the ring-ring interactions which may take place in the intercalation process, we studied the stacking interactions between the aminoquinoline ring of chloroquine and the nucleotide bases. Models of the form B-C3-Q ( 2 , 3, and 4) were prepared, in which the aminoquinoline Q and the base B are joined
by a trimethylene chain;g such molecules are capable of adopting a folded conformation in which the two aromatic systems interact. The proportion of molecules in the folded conformation will be an indication of the affinity of quinoline for the base. In a previous publicationlo we have thus reported a study of the interaction of aminoquinoline with a purine (adenine) and a pyrimidine (thymine);the extent of interaction with adenine was much greater than that with thymine. This result is in agreement with studies of chloroquine-polynucleotide c ~ m p l e x a t i o n ~ ~ ~ which demonstrated only a weak affinity of the quinoline nucleus for pyrimidine bases. However, the studies of the complexation of chloroquine with polynucleotides did not clearly establish the role of the purine bases, adenine and guanine. An ambiguity arises from the fact that most of these studies were based on measurements of the perturbation “hypochromicity”’l in the uv absorption of chloroquine in the complex, The degree of perturbation depends on the nature of the polynucleotide, being stronger for polyG than for polyA, for polydGdC than for polydAdT, and increasing with the (G-C) content in a series of natural DNA’s from different sources, while the equilibrium
DNA Complexes with Antimalarial Compounds
constants are only slightly dependent of the nature of the base. According to some authors,5b the difference observed in the spectrum of chloroquine in the presence of bases is not related to difference in binding; the two bases interact approximately equally with the drug, but guanine exhibits a greater perturbation effect on the chloroquine spectrum as compared with adenine, while, according to the difference reflects a specificity for guanine in the binding. Thus the question remains whether chloroquine shows a specific affinity for guanine or whether for purines in general. In order to study the ability of monomeric adenine and guanine to complex with the aromatic ring of chloroquine and measure the corresponding perturbations in the uv spectra, we have prepared the model systems 3 and 4 in which the quinoline is linked to adenine and guanine, respectively. The spectroscopic properties of these compounds have been compared with those of the reference substances 5 , 6, and 7. We first describe the synthesis of compounds 3-7 and then report a study of
Journal of Medicinal Chemistry, 1977, Vol. 20, No. 1
107
Scheme I
+
CI
11
+ Q-C1) (route 2).
CI
fl
: Ade-C.
i
their spectroscopic properties. Synthesis. Adenine Model, Ade-Cs-Q (3). The nucleophilicity of the anion formed at position 9 of adenine in strongly basic media was used to prepare the 9-alkyl derivatives. The Ade-Cs-Q system could thus be prepared by two different routes. The first one involved bromide displacement from 4-(3-bromopropylamino)-7-chloroquinoline (Q-Cs-Br + Ad-) (route 1). In the latter, the adenine anion was used to prepare the intermediate 9(3-aminopropy1)adenine (Ade-CsNH2) which was then allowed to react with 4,7-dichloroquinoline (Ade-Cs-NHz
A synthesis of the latter type has already been employed by Leonard12J3to prepare coenzyme and dinucleotide (Ade-C3-Ar) models. Route 1 (Scheme I). Reaction of excess 3-aminopropanol with 4,7-dichloroquinoline (130 "C, 12 h) gave 4-(3-hydroxypropylamino)-7-chloroquinoline(9) in 75% yield. The hydroxyl band appears at 3370 cm-l in the ir. The NMR spectrum shows signals characteristic of the aromatic system: -OH at 4.1 ppm, a triplet a t 4.58 ppm due to the hydroxyl-bearing methylene, and a signal at 3.55 ppm due to the methylene group adjacent to the amino function, coupled to the amino proton. Treatment of 9 with 48% HBr at 120 "C gave 443bromopropylamino)-7-chloroquinolinehydrobromide (yield 50%) which was converted to the free base 10 (Q-Cs-Br) using an ion-exchange resin. The NMR of the hydrobromide shows a broad signal a t 9.5 ppm for the -"+. The bromomethylene protons and the methylene protons adjacent to the amino group appear together at 3.7 ppm. Reaction of adenine with sodium hydride in dimethylformamide a t room temperature followed by the addition of the bromo compound 10 gave the desired 4- [3-(aden-9-yl)propylamino]-7-chloroquinoline (3, AdeC3-Q) in 9% yield. The ir spectrum shows bands characteristic of the adenine (1680 and 1650 cm-l) and 4aminoquinoline (3300 and 1590 cm-l) systems. The NMR spectrum (Me2SO) shows signals characteristic of the unprotonated quinoline present in the Q-Cs-X system: two sharps peaks near 8 ppm due to the H-2 and H-8 protons of adenine and the methylene adjacent to the adenine ring appears at 4.26 ppm as in 9-propyladenine. A second product, isolated in 21% yield, was identified (no NH in the ir spectrum; NMR, four equivalent protons geminal to nitrogen, appearing as a triplet a t 4.32 ppm) (11) arising from as the 4-(azetid-l-yl)-7-chloroquinoline an intramolecular reaction.14 It is interesting that a similar intramolecular reaction of the homologue 4-bromobutylaminoquinoline, the precursor of a large number of
108 Journal of Medicinal Chemistry, 1977, Vol. 20, No. 1 Scheme 11
Bolte, Demuynck, Lhomme
Scheme I11 “2
0
0
chloroquine analogues,15has never been observed. Because of the strong allergenic properties of 10 noted during this work, the second route was developed as an alternative synthesis. Route 2 (Scheme 11). Reaction of 4,7-dichloroin dimethyl quinoline with 9-(3-arnin0propyl)adenine~~ sulfoxide at 110 “ C for 14 h gave the desired 4-[3(aden-9-yl)propylamino]-7-chloroquinoline (3, Ade-Cs-Q) in 60% yield. The use of dimethyl sulfoxide considerably increased the efficiency of the reaction, compared to hydroxylic solvents (ethoxyethanol, for example) generally used in this series as reaction medium.16 Guanine Model, Gua-Cs-Q (4). Direct alkylation of guanine at the 9 position poses more problems than that of adenine. Firstly, prior protection of the 2-amino group is necessary. Secondly, the alkylation results in a mixture of the 7- and 9-derivatives.’; For these reasons, it is preferable to carry out the alkylation on a precursor of the guanine system. 2-Amino-4-hydroxy-5-nitro-6-chloropyrimidine (13)l8 serves this purpose, requiring subsequent amination by the proper 4-(3-aminopropylamino)-7chloroquinoline (14). The latter compound was obtained by reaction of 4,7dichloroquinoline with a large excess of 1,3-diaminopropane (Scheme 111). The ir spectrum of the hydrochloride of 14 shows absorptions due to a primary amine salt and the quinoline ring system at 3220,1620, and 1590 cm-‘. The NMR shows signals corresponding to an aminoquinoline; the methylene protons adjacent to the primary amine appear at 2.90 ppm. The chloropyrimidine 13 was then allowed to react with the latter 14 to give intermediate 15. The reaction of 13 with 14 with methanol as solvent and triethylamine as base gave low yields due to the limited solubility of the pyrimidine. Using dimethyl sulfoxide and diazabicyclooctane a t room temperature, the yield was much improved, the product precipitating as it formed. The ir shows bands due to the amino group (3380 cm-l), quinoline and pyrimidine systems (1610 and 1660 cm-I), and the nitro group (1570 cm I). The NMR (CF3C02D) shows signals characteristic of the quinoline system; the methylene groups adjacent to the quinoline and pyrimidine rings appear as a broad signal at 3.8 ppm. The nitro group was reduced using zinc in formic acid and then, without isolation of the intermediate amine, the formic acid was replaced with dimethylformamide; heating this solution with potassium carbonate brought about cyclization to the imidazole. The product (Gua-C3-Q)was isolated as the hydrochloride. The ir spectrum shows bands at 3300 (amine) and 1670 and 1610 cm (purine and quinoline rings). The NMR (CF3C02D) shows signals typical of protonated quinoline, a sharp signal at 9.03 ppm due to H-8 of the guanine and a multiplet at 4.63 ppm due to the methylene group adjacent to the guanine ring. The uv spectrum in ethanol is the sum of the spectra of the two chromophores. The
* I
‘CH2J3
CI
.Gr” ‘
ICH2l3
1 H C o o ~ / ~ n t -
z
I
OMf/CO,N.,
N d
C
4_
1 4 ,
’,5
reference compound, 9-propylguanine (7, Gua-C3), was prepared by a similar method.12a Physical-Chemical Study. Method. The conformational study of the model systems allows a determination of the affinity of the quinoline nucleus for adenine and guanine. Schematically, there are two types of conformation: in one the purine base and the quinoline are sufficiently distant from one another that there is no interaction between them (“free” or “open” form); in the other the two aromatic rings are in contact (“folded” or “stacked” form). The uv spectrum of the quinoline nucleus and the purine base should be perturbed in the complexed form as it is in the case of the polynucleotide-chloroquine c ~ m p l e x , ~ and comparison of the model systems B-C3-Q with the reference substances B-C3 and Q-C3 should allow a measurement of this perturbation. To conform with the conditions used in the study of the real system, water was selected here as the solvent. Because of limited solubility, and to avoid intermolecular interactions, very dilute solutions must be used, conditions suited to uv spectroscopy. Results In a preceding publication10we have shown that the uv spectrum of the model system Ad-C3-Q exhibited strong perturbations, being mainly characterized by a strong decrease in the absorption intensity (“hypochromism”); this was interpreted as a proof of the existence of strong interaction between the two chromophores. This hypochromic effect is indeed observed for systems in which the chromophores are stacked one on another. DNA, RNA, synthetic polynucleotides, and dinucleotides all present a hypochromic effect in the uv.19 According to TinocoZ0 and Rhodes,21the hypochromism is due to the electric field created by the electrons of one chromophore acting on the electronic transition moment of the opposite (absorbing) chromophore. The hypochromic effect is defined by 70 H = [l -f(B-Cs-Q)/V(B-Cd f(Q-C3)]]100,where f is the oscillator strength of the transition, i.e., a measure of the intensity of absorption f = 4.32 X 10-gJ[t(X)/X2]dXwhere t is the molecular extinction coefficient. Uv spectra of the model compounds 3 (Ade-Ca-Q) and 4 (Gua-C3-Q) and of the reference compounds 6 (Ade-W, 7 (Gua-Ca), and 5 (Q-C3) were measured, at pH 7,25 “C, and approximately 10-5 M, in water.22 As seen in Figure 1, the spectra of the two model systems 3 and 4 exhibit a little shift of the absorption bands to longer wavelength
+
D N A Complexes with Antimalarial Compounds
Journal of Medicinal Chemistry, 1977, Vol. 20, No. 1
Figure 1. Comparative ultraviolet spectra of model systems (- - -) and reference compounds (-) in water, pH 6.9, 5 x M, 25 "C: (left) interaction of adenine-quinoline, (right) interaction of guanine-quinoline.
Table I. Computed Hypochromism Values ( % H ) for the Base-Quinoline Interaction Models, in Water, 20 " C , pH 7, for the 230-300- and 300-380-nm Range Interaction models Ade-C,-Q ( 3 ) Gua-C,-Q (4) Thy-C,-Q ( 2 )
230-300 nm
300-380
20+ 2 16i 2 5i 2
25.5 * 0.5 25.9 i 0.7 1oi 1
109
lo-'
%n
nm
and a very strong hypochromic effect. A quite similar perturbation is observed for chloroquine in its interaction with DNA.4e This means that the molecules exist essentially under a stacked conformation and, hence, prove the affinity of the quinoline nucleus for both purine bases. In addition, it appears that the calculated H values are very close for the two models23 (Table I, H = 25.5 and 26%). Is this result indicative of equal affinities of chloroquine for adenine and guanine or, in other words, can hypochromic effects induced by adenine and guanine be compared directly? Very few theoretical studies of the hypochromic effect have been done and do not allow a rigorous answer to this question. Nevertheless, the hypochromic effect has been measured for a large number of biological and analogous and the conclusions agree well with results obtained from other sources,24so that hypochromism can be considered to be directly related to the degree of overlap by stacking of two chromophores and constitutes therefore a semiquantitative measurement of the affinity of two chromophores. In the systems we have studied, the results seem to show that the quinoline nucleus has equal affinity for guanine and adenine. However, the measurement of the hypochromic effect does not permit the detection of small differences, nor is it clear that the measured values indicate a strong or moderate quinoline-base interaction. T o answer this question it is necessary to analyze the folding-unfolding process for the two models on a more quantitative basis. One convenient approach, although constituting a crude appro~imation,~5 is to use a two-stage model; the unstacking process is assumed to be a two-state process in which a completely stacked form is in equi-
0
10
20
30
40
SO
60
70
80
TEMP'
Figure 2. Variation of hypochromism values (% H ) with temperature for model system 3 , Ade-C,-Q, in water, pH 5.5.
librium with a totally unstacked conformation. This requires the knowledge of the spectroscopic properties of the two forms. While the optical properties of the unstacked form are known from the reference compounds (0% hypochromism), the problem lies in the determination of the hypochromic effect of the totally stacked conformer. One possible approach is to study the B-C3-Q systems as a function of temperature; if the passage from the open to folded conformation is exothermic, lowering the temperature should increase the latter form. We have measured the hypochromic effect in the region 300-380 nm, where the base does not absorb, for the compounds 3 (Ade-Cs-Q) and 4 (Gua-Ca-Q) as a function of temperature. Measurements were made using sodium phosphate solutions (pH 5.5 a t 20 "C) of the compounds at temperatures from -5 to +80 0C.26 The results are shown in Figures 2 and 3. For Ade-C3-Q, the hypochromic effect is constant from -5 to +15 "C and equal to 26%. This value therefore
110 Journal of Medicinal ('hemistr), 1977. Vol. 20, .Yo. 1
t
l
o
10
20
30
bo
50
60
70
80
TEMP
Figure 3 . Variation of hypochromism values (% H ) with temperature for model system 4, Gua-C;Q, in water, pH 5 . 5 .
Table 11. Variation of Stacking Interaction with Temperature. Percent of Folded Conformation, Measured for the Three Models, in Water at pH 5.5
Ade-C,-Q ( 3 ) , H,, - 26 Gua-C,-Q (4), H m , - 23 5 Thy-C,-Q 14 G H,,, ( 2 ) ,G" 26
100- 4
88. 4
75. 4
100- 1
921 1
83- 1
40
